View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Am. J. Hum. Genet. 70:1568–1574, 2002 Report

The Encoding Nicastrin, a Major g-Secretase Component, Modifies Risk for Familial Early-Onset Alzheimer Disease in a Dutch Population-Based Sample Bart Dermaut,1 Jessie Theuns,1 Kristel Sleegers,2 Hiroshi Hasegawa,3 Marleen Van den Broeck,1 Krist’l Vennekens,1 Ellen Corsmit,1 Peter St. George-Hyslop,3 Marc Cruts,1 Cornelia M. van Duijn,1,2 and Christine Van Broeckhoven1 1Department of Molecular Genetics, Flanders Interuniversity Institute of Biotechnology, University of Antwerp (Universitaire Instelling Antwerpen), Antwerp; 2Department of Epidemiology and Biostatistics, Erasmus Medical Center, Rotterdam; and 3Centre for Research in Neurodegenerative Diseases and Department of Medicine, University of Toronto, Department of Medicine (Neurology), Toronto Western Hospital, University of Toronto, Toronto

Nicastrin regulates g-secretase cleavage of the amyloid precursor by forming complexes with , in which most mutations causing familial early-onset Alzheimer disease (EOAD) have been found. The gene encoding nicastrin (NCSTN) maps to 1q23, a region that has been linked and associated with late-onset Alzheimer disease (LOAD) in various genome screens. In 78 familial EOAD cases, we found 14 NCSTN single-nucleotide polymor- phisms (SNPs): 10 intronic SNPs, 3 silent mutations, and 1 missense mutation (N417Y). N417Y is unlikely to be pathogenic, since it did not alter amyloid b secretion in an in vitro assay and its frequency was similar in case and control subjects. However, SNP haplotype estimation in two population-based series of Dutch patients with EOAD (n p 116 ) and LOAD (n p 240 ) indicated that the frequency of one SNP haplotype (HapB) was higher in the group with familial EOAD (7%), compared with the LOAD group (3%) and control group (3%). In patients with familial EOAD without the APOE e4 allele, the HapB frequency further increased, to 14%, resulting in a fourfold increased risk (odds ratio p 4.1; 95% confidence interval 1.2–13.3;P p .01 ). These results are compatible with an important role of g-secretase dysfunction in the etiology of familial EOAD.

The amyloid b peptide (Ab) is the main component of causative EOAD mutations directly influence g-secretase the neuritic plaque, the histological hallmark lesion in activity, at the level of either the substrate (APP muta- the brains of patients with Alzheimer disease (AD [MIM tions at the g cleavage site) or the (all PSEN 104300]). Ab is generated by b- and subsequent g-se- mutations), g-secretase dysfunction is crucial in familial cretase cleavage of the amyloid precursor protein (APP EOAD pathogenesis. Recent findings by us (van Duijn [MIM 104760]). Mutations in APP and the presenilins et al. 1999; Theuns et al. 2000) and others (Lambert et (PSEN1 [MIM 104311] and PSEN2 [MIM 600759]) al. 2001), of genetic association between the promot- cause the rare familial early-onset type of AD (EOAD) er region of PSEN1 and EOAD, further emphasize the (AD Mutation Database). Since the large majority of importance of altered g-secretase activity in EOAD. Recently, a novel type 1 transmembrane glycoprotein, Received January 9, 2002; accepted for publication March 11, 2002; designated “nicastrin” (NCSTN [MIM 605254]), was electronically published April 24, 2002. identified as a component of the -containing Address for correspondence and reprints: Dr. Christine Van Broeck- g-secretase complex that is involved in the cleaving of hoven, Department of Molecular Genetics (VIB8), University of An- APP and Notch in their transmembrane domains (Yu et twerp (UIA), Universiteitsplein 1, B-2610, Antwerpen, Belgium. E- al. 2000; Chen et al. 2001). Like all pathogenic PSEN mail: [email protected] ᭧ 2002 by The American Society of Human Genetics. All rights reserved. mutations, missense mutations in a conserved hydro- 0002-9297/2002/7006-0019$15.00 philic domain (DYIGSrAAIGS) in NCSTN strongly in-

1568 Reports 1569

Figure 1 Genomic structure of NCSTN and major SNP haplotypes, with their estimated frequencies in 78 patients with familial EOAD. SNPs used for case-control association analysis are in boldface.

creased Ab production, whereas deletions in this domain in 78 patients was classified as familial (i.e., at least one inhibited Ab production (Yu et al. 2000). NCSTN maps first-degree relative with dementia), 10 cases of which to 1q23, a region that has shown evidence for linkage fulfilled our criteria of autosomal dominant inheritance to (Kehoe et al. 1999) and association with LOAD (Hil- (Cruts et al. 1998). In this sample, seven patients were tunen et al. 2001). identified who carried a causative PSEN mutation (five We evaluated the contribution of genetic variations in in PSEN1 and two in PSEN2) (Cruts et al. 1998; authors’ NCSTN in two large series of patients with EOAD (onset unpublished data). Patients with EOAD were compared before or at age 65 years) and LOAD (onset after age with a control group of 114 individuals in the same age years; 46% 3.3 ע years). Patients with EOAD were derived from a range (mean age at examination60.6 65 Dutch population-based study of EOAD in the four male). Patients with LOAD were drawn from the Rot- northern provinces of the Netherlands and metropolitan terdam study, a population-based prospective study on Rotterdam (van Duijn et al. 1994a). The patients were residents of age у55 years from a Rotterdam suburb in sampled during two study periods. The original sample the Netherlands (Hofman et al. 1991). Dementia was was collected between 1980 and 1987 and has been diagnosed by a three-step examination, as detailed else- described elsewhere (van Duijn et al. 1994b). The initial where (Ott et al. 1995). The diagnosis of probable AD study was extended between 1997 and 2000, in a ge- was based on NINCDS-ADRDA criteria (McKhann et netically isolated part of the previously described area, al. 1984): 339 patients with AD were ascertained at the with the same sampling criteria. The current sample start of the study, and 116 patients with incident AD comprises 96 patients from the original group, with an were diagnosed at the first follow-up period. DNA was additional 20 patients from the extended study. Mean available for 240 patients with probable LOAD cases -years; 20% male). Pa 6.9 ע years, and 23% of patients (mean age at onset84.1 5.5 ע age at onset was56.4 in the sample were male. The diagnosis was indepen- tients with LOAD were compared to 239 control indi- ;years 3.7 ע dently confirmed by a member of the research team and viduals (mean age at examination72.1 by a neurologist using a standardized protocol consistent 15% male). All control individuals were taken from the with the National Institute of Neurological and Com- Rotterdam study and were screened for cognitive dys- municative Disorders and Stroke-Alzheimer’s Disease function. Potential control subjects with dementia were and Related Disorders Association (NINCDS-ADRDA) excluded. criteria for AD (McKhann et al. 1984). The disease in We determined the genomic organization of NCSTN 38 patients was classified as sporadic, and the disease (gDNA: 15.9 kb; cDNA: 2.9 kb) by alignment of the 1570 Am. J. Hum. Genet. 70:1568–1574, 2002 complete cDNA sequence (NCSTN cDNA [GenBank Table 1 accession number AF240468]) against the human work- NCSTN Sequence Variations and Their Allele Frequencies in 78 ing draft sequence ( Project Work- Familial EOAD Cases ing Draft Web site). NCSTN consists of 17 exons (fig. No. (Frequency) 1), all of which contain coding sequence. Using PCR Location Polymorphism of Mutant Allelea primers flanking each exon, we screened NCSTN exons Exon 3 c.237GrA (Glu79) 2 (.01) 1–16 and the coding part of exon 17 by denaturing high- Intron 4 IVS4Ϫ69CrG 1 (.01) performance liquid chromatography (DHPLC) in 78 pa- Intron 5 IVS5Ϫ74CrT 13 (.08) tients with familial AD, including 10 patients with au- Exon 6 c.636ArG (Leu212) 13 (.08) tosomal dominant EOAD. Direct sequencing of PCR Intron 6 IVS6ϩ18CrG 8 (.05) r fragments with aberrant DHPLC patterns revealed 14 Exon 7 c.747C T (Asp249) 13 (.08) Intron 7 IVS7ϩ61GrA 2 (.01) single nucleotide polymorphisms (SNPs), of which 10 Intron 10 IVS10ϩ32CrG 1 (.01) were located in introns and 4 in coding regions (table Intron 10 IVS10Ϫ5CrG 13 (.08) 1). Direct sequencing of NCSTN in the probands of the Exon 11 c.1249ArT (Asn417Tyr) 1 (.01) 10 autosomal dominant families revealed no causative Intron 13 IVS13Ϫ129TrC 16 (.10) r Intron 15 IVS15ϩ31insG 1 (.01) mutations or additional variations. Only c.1249A T Ϫ r r Intron 16 IVS16 119G C 16 (.10) predicts an amino acid substitution (N Y at codon 417) Intron 16 IVS16Ϫ74CrT 3 (.02) and was found in a patient with familial EOAD (age at a p onset 57 years). No other family members were available Frequencies are given as proportion ofN 156 total alleles. for segregation analysis. NCSTN N417Y is a conser- vative amino acid substitution affecting a residue that is not evolutionarily conserved in mouse (A), Drosophila as “HapA,” “HapB,” “HapC,” and “HapD”) (fig. 1). (E), or Caenorhabditis elegans (G) (Yu et al. 2000) but Through use of a pyrosequencing assay, genotypes for is located within a putative glycosylation site (N-Q-S). each of these four SNPs were determined in patients To test the potential contribution of N417Y to AD, we with EOAD, patients with LOAD, and with their age- used a pyrosequencing assay (Alderborn et al. 2000; matched control groups (table 2). Genotype frequencies Theuns et al. 2001) to screen all patients with EOAD, for each SNP were in Hardy-Weinberg equilibrium as well as 235 patients with probable LOAD and a total (HWE) in all case and control groups (P 1 .23 ). No sta- of 474 control subjects taken from the Rotterdam study. tistically significant differences in genotype frequencies In addition to the patient with familial EOAD, N417Y were observed between patients with EOAD and LOAD was also found in one patient with LOAD (familial; age and control subjects, at any of the four SNP loci tested. at onset 87 years) and in three control subjects without We then estimated four-locus haplotype frequencies dementia (ages 84, 87, and 95 years), resulting in iden- from genotype data with a maximum-likelihood method tical N417Y carrier frequencies in patients with AD using the expectation-maximization algorithm under the (2/354 p 0.6% ) and controls ( 3/474 p 0.6% ). To as- assumption of HWE (Excoffier and Slatkin 1995). Es- sess whether NCSTN N417Y influences APP processing, timated haplotype distributions in the different groups we measured Ab40 and Ab42 levels in conditioned me- were compared using CLUMP (Sham and Curtis 1995). dia of HEK-293 cells transfected with the mutant and CLUMP generates empirical P values that are based on wild-type NCSTN cDNA. Results were compared with Monte Carlo simulations and is useful for case-control PSEN1 L392V, a clinical mutation causing familial comparisons of multiallelic genetic markers. To test for EOAD that is known to increase Ab42 levels (Sherring- interaction between NCSTN and APOE, the case and ton et al. 1995; Citron et al. 1997), and PSEN1 D385A, control series were stratified for the presence of an APOE an artificial mutation that substantially reduces Ab pro- e4 allele. Haplotypes were estimated separately in the duction (Wolfe et al. 1999). Although both PSEN1 mu- patients with familial EOAD. Because haplotype fre- tations showed the expected effects on Ab production, quencies were not statistically different between the two Ab40 and Ab42 levels were similar for N417Y and wild- control groups (P p .48 ), and in order to obtain reliable type NCSTN (data not shown). haplotype frequency estimates in the different APOE To further evaluate the possibility that variations in strata, both control groups were pooled. In all groups NCSTN affect susceptibility to EOAD or LOAD, case- studied, haplotype estimation revealed one frequent control association analysis was performed. Because of haplotype (HapA) and three rarer haplotypes (HapB, strong linkage disequilibrium between the SNPs in the HapC, and HapD) (table 3). We found no significant 78 familial EOAD samples, we selected four informative differences in NCSTN haplotype distributions between SNPs (c.636ArG, IVS6ϩ18CrG, IVS10Ϫ5CrG, and patients with LOAD and control subjects (table 3). IVS16Ϫ119GrC) that allowed definition of four ma- Moreover, haplotype frequencies remained highly stable jor haplotypes: A, B, C, and D (hereafter referred to across APOE strata in both the LOAD and control Reports 1571

Table 2 Genotype Frequencies of Four Selected NCSTN SNPs in Patients with EOAD, Patients with LOAD, and Control Subjects

EOAD VERSUS CONTROL GROUP LOAD VERSUS CONTROL GROUP No. (Frequency) in No. (Frequency) in SNP AND GENOTYPE EOAD Group Control GroupP a LOAD Group Control Group P a c.636ArG: AA 103 (.89) 102 (.89) 220 (.94) 175 (.93) .87 .85 AG 13 (.11) 12 (.11) 14 (.06) 12 (.07) Total 116 114 234 187 IVS6ϩ18CrG: CC 108 (.93) 102 (.89) 217 (.92) 198 (.91) .33 b .90 CGϩGG 8 (.07) 12 (.11) 20 (.08) 19 (.09) Total 116 114 237 217 IVS10Ϫ5CrG: CC 102 (.88) 102 (.89) 215 (.92) 219 (.94) .71 .48 CG 14 (.12) 12 (.11) 19 (.08) 15 (.06) Total 116 114 234 234 IVS16Ϫ119GrC: GG 99 (.85) 97 (.85) 188 (.88) 167 (.92) .96 c .15 GCϩCC 17 (.15) 17 (.15) 26 (.12) 14 (.08) Total 116 114 214 181 a Calculated using the x2 statistic with 1 df. b Includes one subject with the GG genotype. c Includes two subjects with the CC genotype.

groups, suggesting no interaction between NCSTN and mutation screening in patients with familial EOAD and APOE in the risk to develop LOAD. In the patients with subsequent case-control association analysis using four EOAD, however, we found an increased frequency of NCSTN polymorphisms in large population-based series HapB in those with familial EOAD (7%) and APOE of EOAD and LOAD. Although screening of all NCSTN e4(Ϫ) EOAD (7%), compared with patients with LOAD coding exons and flanking intronic sequences in 78 pa- (3%) and control subjects (3%). In patients with familial tients with familial EOAD revealed a total of 14 NCSTN EOAD without APOE e4 alleles, the HapB frequency sequence variations, only NCSTN N417Y,detected in one further increased, to 14%, resulting in a significantly patient with familial EOAD, was potentially pathogenic. different haplotype distribution (P p .01 ). When cal- However, its equal frequency in AD cases and controls culating risks by comparing the HapB frequencies in case (allele frequency 0.3%), its inability to alter Ab produc- subjects and their age-matched control subjects and by tion in an in vitro assay, and its low evolutionary con- using the non-HapB haplotypes as reference, we found servation in different species strongly suggest that N417Y that the HapB frequency was significantly increased in is a very rare but nonpathogenic variant. Although we patients with familial EOAD without APOE e4 alleles can therefore conclude that mutation of NCSTN is not (odds ratio [OR] 4.1; 95% CI 1.2–13.3;P p .01 ). No a frequent cause of EOAD, genetic association analy- homozygotes for HapB were observed. Assuming that sis did reveal that the risk of developing familial EOAD individuals heterozygous at c.636ArG, IVS10Ϫ5CrG, was modified by a significantly overrepresented NCSTN and IVS16Ϫ119GrC are heterozygous HapB carriers, haplotype (HapB) in patients with familial EOAD (7%), we found a significantly increased HapB/Hap* (where particularly in those who lack APOE e4 alleles (14%), Hap* is any non-HapB haplotype) genotype frequency compared with control subjects (3%) and patients with in patients with familial APOE e4(Ϫ) EOAD (33%), LOAD (3%). Although we acknowledge the possibility compared with age-matched APOE e4(Ϫ) control sub- of a false-positive finding, from a biological point of view jects (9%) (OR 4.9, 95% CI 1.5–16.3;P p .005 ). Since the association between NCSTN and familial EOAD none of the apparent HapB carriers with EOAD had a seems highly plausible, for several reasons. First, the as- PSEN mutation, the exclusion of the seven PSEN mu- sociation is caused by a specific overrepresentation of one tation carriers did not alter these results. haplotype in patients and is not due to frequency shifts In the present study, we have investigated the contri- in the control group, where NCSTN haplotypes remained bution of genetic variations in NCSTN to the occurrence highly stable even when stratified for APOE. Second, it of AD, through use of a combined strategy of extensive is of interest that the most dramatic increase of HapB was 1572 Am. J. Hum. Genet. 70:1568–1574, 2002

Table 3 Estimated NCSTN Four-Locus Haplotype Frequencies in Patients with EOAD, Patients with Familial EOAD, Patients with LOAD, and Control Subjects, Stratified for the Presence of an APOE e4 Allele

CONTROL LOAD GROUP EOAD GROUP FAMILIAL EOAD GROUP SAMPLE AND GROUP HAPLOTYPEa FREQUENCY Frequency x2 P b Frequency x2 P b Frequency x2 P b Total Sample:c A (ACCG) .89 .89 .89 .85 B(GCGC) .03 .03 .05 .07 C(AGCG) .04 .04 1.1 .91 .03 2.3 .69 .05 7.8 .10 D (ACCC) .02 .02 .02 .02 Other .02 .02 .01 .01 APOE e4(ϩ):d A (ACCG) .92 .88 .92 .89 B(GCGC) .03 .01 .03 .05 C(AGCG) .04 .07 4.4 .35 .03 .1 1.00 .05 .4 .98 D (ACCC) .01 .01 .01 .01 Other .01 .03 .01 .01 APOE e4(Ϫ):e A (ACCG) .89 .89 .82 .74 B(GCGC) .03 .03 .07 .14 C(AGCG) .04 .02 4.3 .37 .01 8.9 .06 .05 15.1 .01 D (ACCC) .02 .03 .05 .05 Other .02 .02 .04 .03 a Underlined bases are those that differ from the wild-type haplotype. b Generated using CLUMP, by random simulation of 1,000 2#5 tables with the same marginal totals as the one under consideration and by counting the number of times that a x2 value associated with the real table was achieved. c Control group,n p 472 ; LOAD group,n p 426 ; EOAD group,n p 232 ; familial EOAD group,n p 156 . d Control group,n p 130 ; LOAD group,n p 160 ; EOAD group,n p 152 ; familial EOAD group,n p 114 . e Control group,n p 342 ; LOAD group,n p 266 ; EOAD group,n p 80 ; familial EOAD group,n p 42 . observed in patients with familial EOAD who lack all with HapB that influences nicastrin expression levels well-established genetic factors (i.e., APOE e4 alleles and might explain our association. In addition, it cannot be PSEN or APP mutations), suggesting that the familial oc- excluded that variations in a nearby gene in LD with currence within this more homogeneous subgroup might NCSTN HapB may be responsible for the observed as- at least be partially attributed to the NCSTN locus. Our sociation. According to the human draft sequence data- results also suggest that an association in the total EOAD base, NCSTN is flanked by the gene encoding the a sub- group may have been masked by the high number of unit of the coatamer (COPA [MIM APOE e4 carriers (65% of the sample), while the effect 601924]) at its 5 side and by the gene encoding the nes- is only present in non-e4 carriers. Third, our finding of cient helix loop helix 1 protein (NHLH1 [MIM 601924]) an effect in EOAD but not in LOAD is in line with our at the 3 side. However, there is currently no direct evi- previous studies showing an association of the PSEN1 dence that these are implicated in AD pathogenesis. regulatory region locus with EOAD (van Duijn et al. In conclusion, our data show that mutation of the 1999; Theuns et al. 2000) but not LOAD (Dermaut et NCSTN coding region is not a frequent cause of EOAD al. 2001) and is consistent with the concept that an in- but suggest that a specific NCSTN haplotype modifies creased production of Ab by altered g-secretase activity the risk of developing familial EOAD, particularly in is the primary pathogenetic event in (familial) EOAD. In those cases that lack an APOE e4 allele. Contrary to our contrast, in LOAD, the relative importance of altered pro- expectation that NCSTN might be responsible for the duction versus degradation or clearance of Ab is still a reported positional evidence of an LOAD gene on 1q23, matter of debate. Although independent replication stud- variation in NCSTN does not confer risk of LOAD in ies are needed to further validate our findings, it is pres- our sample. ently unclear what the underlying biological mechanism of the association might be. Because we detected only intronic or silent variations, we are currently screening Acknowledgments the upstream regulatory region of NCSTN, in order to We are grateful to Jurgen Del-Favero, Hubert Backhovens, assess the possibility that an undetected variation in LD Dirk Van den Bossche, and Sam Sluijs, for their skilled tech- Reports 1573 nical assistance in the mutation screening and genotyping. We Van Broeckhoven C, van Duijn CM (2001) Variable ex- thank Jeanette Vergeer, Wilma Luiten, and Bianca de Graaf, pression of presenilin 1 is not a major determinant of risk for their help in the laboratory analysis, and Gerwin Roks, for late-onset Alzheimer’s Disease. J Neurol 248:935–939 for the sampling of patients. Financial support was received Excoffier L, Slatkin M (1995) Maximum-likelihood estimation from the Fund for Scientific Research Flanders, Federal Office of molecular haplotype frequencies in a diploid population. for Scientific, Technical, and Cultural Affairs Interuniversity Mol Biol Evol 12:921–927 Attractionpoles, the International Alzheimer’s Research Foun- Hiltunen M, Mannermaa A, Thompson D, Easton D, Pirska- dation, the Alzheimer Association, the University of Antwerp nen M, Helisalmi S, Koivisto AM, Lehtovirta M, Ryynanen (UIA), and by grants from the Netherlands Organisation for M, Soininen H (2001) Genome-wide linkage disequilibrium Scientific Research, The Netherlands Institute of Health Sci- mapping of late-onset Alzheimer’s disease in Finland. Neu- ences, the Hersenstichting Nederland, the Municipality of Rot- rology 57:1663–1668 terdam, the Canadian Institutes of Health Research, Howard Hofman A, Grobbee DE, de Jong PT, van-den Ouweland FA Hughes Medical Institute, and the Alzheimer Society of On- (1991) Determinants of disease and disability in the elderly: tario. M.C. and J.T. are postdoctoral fellows—and B.D. is a the Rotterdam Elderly Study. Eur J Epidemiol 7:403–422 Ph.D. fellow—of the Fund for Scientific Research Flanders. Kehoe P, Wavrant-De Vrieze F, Crook R, Wu WS, Holmans P, Fenton I, Spurlock G, Norton N, Williams H, Williams N, Electronic-Database Information Lovestone S, Perez-Tur J, Hutton M, Chartier-Harlin MC, Shears S, Roehl K, Booth J, Van Voorst W, Ramic D, Wil- Accession numbers and URLs for data in this article are as liams J, Goate A, Hardy J, Owen MJ (1999) A full genome follows: scan for late onset Alzheimer’s disease. Hum Mol Genet 8: 237–245 AD Mutation Database, http://molgen-www.uia.ac.be/ Lambert JC, Mann DM, Harris JM, Chartier-Harlin MC, ADMutations/ Cumming A, Coates J, Lemmon H, StClair D, Iwatsubo T, GenBank, http://www.ncbi.nlm.nih.gov/Genbank/ (for hu- Lendon C (2001) The Ϫ48 C/T polymorphism in the pre- man NCSTN [accession number AF240468]) senilin 1 promoter is associated with an increased risk of Human Genome Project Working Draft, http://genome.ucsc developing Alzheimer’s disease and an increased Ab load in .edu/ brain. J Med Genet 38:353–355 Online Mendelian Inheritance in Man (OMIM), http://www McKhann G, Drachman D, Folstein M, Katzman R, Price D, .ncbi.nlm.nih.gov/Omim/ (for AD [MIM 104300], APP Stadlan EM (1984) Clinical diagnosis of Alzheimer’s disease: [MIM 104760], COPA [MIM 601924], PSEN1 [MIM report of the NINCDS-ADRDA Work Group under the aus- 104311], PSEN2 [MIM 600759], NHLH1 [MIM 162360], pices of Department of Health and Human Services Task and NCSTN [MIM 605254]) Force on Alzheimer’s Disease. Neurology 34:939–944 Ott A, Breteler MM, van Harskamp F, Claus JJ, van der Cam- References men T, Grobbee DE, Hofman A (1995) Prevalence of Alz- heimer’s disease and vascular dementia: association with ed- Alderborn A, Kristofferson A, Hammerling U (2000) Deter- ucation. The Rotterdam study. BMJ 310:970–973 mination of single-nucleotide polymorphisms by real-time py- Sham PC, Curtis D (1995) Monte Carlo tests for associations rophosphate DNA sequencing. Genome Res 10:1249–1258 between disease and alleles at highly polymorphic loci. Ann Chen F, Yu G, Arawaka S, Nishimura M, Kawarai T, Yu H, Hum Genet 59:97–105 Tandon A, Supala A, Song YQ, Rogaeva E, Milman P, Sato Sherrington R, Rogaev EI, Liang Y, Rogaeva EA, Levesque G, C, Yu C, Janus C, Lee J, Song L, Zhang L, Fraser PE, St Ikeda M, Chi H, et al (1995) Cloning of a gene bearing George-Hyslop PH (2001) Nicastrin binds to membrane- missense mutations in early-onset familial Alzheimer’s dis- tethered Notch. Nat Cell Biol 3:751–754 ease. Nature 375:754–760 Citron M, Westaway D, Xia W, Carlson G, Diehl T, Levesque Theuns J, Del Favero J, Dermaut B, van Duijn CM, Backhovens G, Johnson WK, Lee M, Seubert P, Davis A, Kholodenko H, van den Broeck M, Serneels S, Corsmit E, Van Broeck- D, Motter R, Sherrington R, Perry B, Yao H, Strome R, hoven C, Cruts M (2000) Genetic variability in the regulatory Lieberburg I, Rommens J, Kim S, Schenk D, Fraser P, St region of presenilin 1 associated with risk for Alzheimer’s George-Hyslop P, Selkoe DJ (1997) Mutant presenilins of disease and variable expression. Hum Mol Genet 9:325–331 Alzheimer’s disease increase production of 42-residue am- Theuns J, Feuk L, Dermaut B, Del Favero J, Roks G, Van yloid beta-protein in both transfected cells and transgenic den BD, Corsmit E, van den BM, van Duijn CM, Cruts M, mice. Nat Med 3:67–72 Brookes AJ, Van Broeckhoven C (2001) The TNFRSF6 Cruts M, van Duijn CM, Backhovens H, van den Broeck M, gene is not implicated in familial early-onset Alzheimer’s Wehnert A, Serneels S, Sherrington R, Hutton M, Hardy J, disease. Hum Genet 108:552–553 St George-Hyslop PH, Hofman A, Van Broeckhoven C (1998) van Duijn CM, Cruts M, Theuns J, Van Gassen G, Backhovens Estimation of the genetic contribution of presenilin-1 and -2 H, van den Broeck M, Wehnert A, Serneels S, Hofman A, mutations in a population based study of presenile Alzheimer Van Broeckhoven C (1999) Genetic association of the pre- disease. Hum Mol Genet 7:43–51 senilin-1 regulatory region with early-onset Alzheimer’s dis- Dermaut B, Roks G, Theuns J, Rademakers R, Houwing-Duis- ease in a population-based sample. Eur J Hum Genet 7: termaat JJ, Serneels S, Hofman A, Breteler MB, Cruts M, 801–806 1574 Am. J. Hum. Genet. 70:1568–1574, 2002 van Duijn CM, de Knijff P, Cruts M, Wehnert A, Havekes LM, Wolfe MS, Xia W, Ostaszewski BL, Diehl TS, Kimberly WT, Hofman A, Van Broeckhoven C (1994a) Apolipoprotein E4 Selkoe DJ (1999) Two transmembrane aspartates in presenil- allele in a population-based study of early-onset Alzheimer’s in-1 required for presenilin endoproteolysis and gamma- disease. Nat Genet 7:74–78 secretase activity. Nature 398:513–517 van Duijn CM, Hendriks L, Farrer LA, Backhovens H, Cruts Yu G, Nishimura M, Arawaka S, Levitan D, Zhang L, Tandon M, Wehnert A, Hofman A, Van BC (1994b) A population- A, Song YQ, et al (2000) Nicastrin modulates presenilin- based study of familial Alzheimer disease: linkage to chro- mediated notch/glp-1 signal transduction and betaAPP pro- mosomes 14, 19, and 21. Am J Hum Genet 55:714–727 cessing. Nature 407:48–54